[PATCH] virt: acrn: stop CPU removal when offlining fails

[PATCH] virt: acrn: stop CPU removal when offlining fails

[Avi Weiss]

remove_cpu_store() ignores errors from remove_cpu() and calls
hcall_sos_remove_cpu() even when Linux cannot offline the CPU.
This can leave Linux and ACRN with inconsistent CPU ownership.

Return the offlining error before issuing the hypercall. If the
hypercall subsequently fails, bring the CPU back online only when
it was online before the removal attempt.

Fixes: 279dcf693ac76 ("virt: acrn: Introduce an interface for Service VM to control vCPU")
Signed-off-by: Avi Weiss <thnkslprpt@gmail.com>
---
Compile-tested CONFIG_ACRN_HSM=m with W=1; no ACRN Service VM runtime test.
 drivers/virt/acrn/hsm.c | 12 +++++++++---
 1 file changed, 9 insertions(+), 3 deletions(-)

diff --git a/drivers/virt/acrn/hsm.c b/drivers/virt/acrn/hsm.c
index f170ff4617fdd..79f8e50f0ac5c 100644
--- a/drivers/virt/acrn/hsm.c
+++ b/drivers/virt/acrn/hsm.c
@@ -437,6 +437,7 @@ static ssize_t remove_cpu_store(struct device *dev,
 				const char *buf, size_t count)
 {
 	u64 cpu, lapicid;
+	bool was_online;
 	int ret;
 
 	if (kstrtoull(buf, 0, &cpu) < 0)
@@ -445,8 +446,12 @@ static ssize_t remove_cpu_store(struct device *dev,
 	if (cpu >= num_possible_cpus() || cpu == 0 || !cpu_is_hotpluggable(cpu))
 		return -EINVAL;
 
-	if (cpu_online(cpu))
-		remove_cpu(cpu);
+	was_online = cpu_online(cpu);
+	if (was_online) {
+		ret = remove_cpu(cpu);
+		if (ret)
+			return ret;
+	}
 
 	lapicid = cpu_data(cpu).topo.apicid;
 	dev_dbg(dev, "Try to remove cpu %lld with lapicid %lld\n", cpu, lapicid);
@@ -459,7 +464,8 @@ static ssize_t remove_cpu_store(struct device *dev,
 	return count;
 
 fail_remove:
-	add_cpu(cpu);
+	if (was_online)
+		add_cpu(cpu);
 	return ret;
 }
 static DEVICE_ATTR_WO(remove_cpu);

base-commit: d89e1d186ec9271bf78e62771779e9fc03bf0fc3
-- 
2.43.0